MESH DEFORMER PROJECT NEEDS YOUR VOICE

[Codewarrior Congrejo]

 i did not say that.

Maybe you wanted to reply to someone else, or just to the thread in general ? =)

And i promise i will leave your avatar alone >.> 

[WhiteRabbit0]

The creation and implementation of the deformer in no way will cause the old avatars to be removed from the grid, or impact them in any way whatsoever.  Your avatar will remain as it was.

 

It will however allow for the creation of beautiful new avatars that run independantly from the default avatar with much better detail and animation which you can chose to purchase.

 

No one will be forcing you to change though unless LL comes down with some reaaaaaaaally bizarre orders out of the blue.

[Gaia Clary]

So this sounds like there is ongoing work on a new Avatar mesh. Where can i read about it ? Detailed information about its features, state of development, estimate of release date, etc. ? Some time ago i also heard (a rumour?) about Linden Lab is working on an improvement on the default avatar regarding its weight maps. So maybe you refer to that ?

[WhiteRabbit0]

The updated Avatar is something that is often requested but officially no work has started on it.  I find it highly highly unlikely that there is any LL project in development currently to do such a thing, however, one of the 100's of possible uses for the deformer is to create said Avatar 2.0,  (3.0 really.. current AV is 2.0, just no one remembers 1.0) .  It would not be an official release that works with the current baked clothing layers, but it would be manageable with clothing appliers and would be done by the community of designers.

 

There is currently another side project that slightly changes the avatar weighting to reduce the horrendous stretching in the current default AV.

 

All progress on that is documented in the Snowstorm 1800 project JIRA, which if you are thoroughly bored, you can read here : https://jira.secondlife.com/browse/STORM-1800?

TLDR version of the JIRA : Slight improvement but might make some current content clip the avatar in areas it did not clip before.  Basically, might break content, probably won't be implemented.

[Coby Foden]

---

WhiteRabbit0 wrote:

The updated Avatar is something that is often requested but officially no work has started on it.

---

Which is a pity.  Due to bad avatar mesh there now starts to be all kinds of body enhancement add-ons to hide the default avatar mesh:

Mesh feet, mesh hands, mesh breasts, complete mesh heads, and even full mesh avatars.

Then we still have the default avatar mesh under all those add-ons.  Even though the avatar mesh, or parts of it, are hidden by alpha mask, isn't it so that the avatar mesh must still be handled by the system?  At least I understand it like that, even if hidden the system must calculate the position of every vertex of the avatar mesh.

Then we have all those add-ons, and thus the system has a lot more calculation to do - avatar mesh plus the add-ons.  If the new avatar mesh was made skillfully avoiding too hig polygon count, still making it good looking in all places, wouldn't that be less laggy than the present style of hiding the avatar mesh and wearing all the possible add-ons?  So I'm wondering, wouldn't it be better if we did had new avatar mesh with more polygons than the current one, eliminating the need to wear all those add-ons?

Just wondering, I'm not sure if my thinking is right or not...

 

[Rahkis Andel]

That'd be nice for human avatars. Not so useful for non-human ones.

A better idea would be to have an "avatar" slot that could be replaced entirely -- no hiding necessary. I don't know how that could be accomplished the way the system is now, though.

[Codewarrior Congrejo]

So I'm wondering, wouldn't it be better if we did had new avatar mesh with more polygons than the current one, eliminating the need to wear all those add-ons?

My personal thought about that, and the whole avatar 2.0 subject:

I do believe that the one of the possibly best solutions for everyone would be to have an option to literally derender /
deactivate the underlying default avatar whenever you choose to wear either so many attachments that is is not visible
anymore or a full body mesh avatar on top of it.

This way we would reduce the cost it still has underneath the things worn on top.
And people could choose to wear whatever mesh avatar they personally like the most. (and on top of that no matter if human or not)

And since it's hard to serve the uphold of existing content by replacing the skeleton (to fix some of the issues it has) it would probably be best to just add the capability to either upload own skeletons or to insert changes to the existing one, if you would like to change something for a certain rigg to work. Possibly even add a limited amount of additional bones.

On top of that I'd wish for the ability to upload a limited amount of own shape keys to deform a full body avatar or attachments based on other very own topology. This would defy the need of the deformer, yes. But bring up the possibility
(along with the other 2 options) to create fully functional replacements for the existing avatar all on our own. 

Because even if secondlife makes a new higher detailed mesh, there still remain certain points to be considered:

- (as stated in one of my former answers and here) - the upholding of existent content / old content remaining usable
- and the fact that we still would have people consider even the new avatar not to be exactly what they wanted.
  And feeling the need to replace certain body parts or even the whole thing with something else.

With these choices, users could either stick with the old avatar as-is, or covered with items to replace certain parts,
or just replace it with the variety of full body meshes / avatars being provided by creators, and choose what ever they personally prefer in terms of its look. 

Surely it would be nice to have for easy solutions to just have a generally 'nicer' looking avatar you could just use as-is again and put some clothes on etc. But I think it would by far not solve all mentioned problems.

I can only hope and wish, though : )

[Coby Foden]

That replacing the avatar mesh with something else sounds excellent.

Linden Lab could provide some human shapes and various creatures of the wildest imagination.  Users could create their own stuff to replace the avatar mesh.  And the present avatar mesh would still be available for the existing content.  Most excellent solution satisfying everybody.

 

[Gaia Clary]

---

Codewarrior Congrejo wrote:

I do believe that the one of the possibly best solutions for everyone would be to have an option to literally derender /

deactivate the underlying default avatar whenever you choose to wear either so many attachments that is is not visible

anymore or a full body mesh avatar on top of it.

---

I thought that the alpha masks have been invented to do just that. I always understood that the alpha mask does not make the Avastar transparent, but tells the render system to not render the masked parts. So if you wear a full transparent mask then you effectively remove the default avatar mesh and only keep the skeleton intact.

what convinces me that this is so ?

 

1. Alpha Masks have only 2 states: Fully visible and fully invisible (translates to "rendered of not rendered")
2. There seems to be no Alpha glitch when wearing Alpha masks. At least i am not aware of it.

However since i do not know for sure if this is true, maybe someone who knows the technical details can approve that i am right. Or correct me if i am wrong ?

thanks.

[Codewarrior Congrejo]

Hey Gaia : )

ok, here is how i understand it;

the wiki (http://wiki.secondlife.com/wiki/Viewerhelp:Appearance_Editor_-_Alpha) doesn't give much info.

And the included link to alpha just says what is already clear to me: 
alpha computer graphics, alpha compositing is the process of combining an image with a background to create the appearance of partial transparency. Textured objects and avatars in Second Life can be rendered partially or completely transparent through the use of alpha masks.

Which would be a pointer towards just regular transparency.
(http://en.wikipedia.org/wiki/Alpha_compositing)

And alpha Images or channels in general are still something that is 'rendered' just rendered as transparent.

Transparency itself is actually pretty cost intense in terms of rendering, due to creation of several drawcalls / figuring what is layered on top of each other, etc. (and increasing- the more transparent layers are overlapping)

The not noticeable alpha / Z sorting issue, is based on the fact that we deal here with a 1 Bit alpha image. Means it accepts only 100% transparent or 100 % opaque (black / white), and is handled by a so called cut-out-shader. Thus those alpha layers are not 'sorted' the same way as other alpha textures.

The fact that it is a 1 Bit Alpha / Cut out shader -  also leads to the following points:  

The fact that the Alpha layer can *hide* the body at 'any given pixel' of the map (so to say) if it would be polygon / faces that are actually derendered or even removed, then the whole face would have to dissapear - when being based on the gemoetry and not just mapped transparency.  (just as it does with whole faces and when a rigged mesh is worn where some verts have no weights and being derendered, or like a mask-modifier in blender behaves)


*** Edit: to show what i meant with that: 


if it actually would remove the faces / geometry one shouldn't be able to :
- make a circle (as on the left side) because it would need to hide all influenced faces and thus i should see 'polygonal
  patterns' missing. The fact i can make a circle leads me to the conclusion its simply layering transparency over it.
- same with the example in the middle:  diagonal cut.

- And on the right side i took it further,  i detached everything (except the things that can't be taken off like the shape)
  put on a full set of 100% alpha layer to hide everything. And it still renders the avatar as textured with 1000.
  Which means the avatar is defenitally still there and not completely derenderd. (including the geometry being still there)

Edit 2 : i am still doing some tests on the 'checkbox' to hide full parts. which is supposed to take off a full part from the baking process. i will return once i have that done. 

>> okay even with fully enabled check-boxes on all bodyparts, the textured 1000 still remains, and so does the shadow the avatar throws onto the ground.
So even here we are just still on the regular 1 Bit Cut out with no removal of topology at all.

Maybe that changes when we have the server side baking in place but so far it at least doesn't seem to do anything in regards of 'really' removing the avatar completely. Maybe anyone has a final answer to this ? I would love to hear that too.

Aside from that (no matter how the avatar would be derendered ) the general  wishes still remain : )

[Rahkis Andel]

I certainly don't have a final answer, but I'll also point out that even with a full body alpha, your avatar's shadow is still there, being calculated in real time.

That is rather telling evidence.

[Codewarrior Congrejo]

about the shadow :  correct, was considering that too, was still doing some tests before jumping to that conclusion.

Updated my former post.

Because now I also have tried to enable all checkboxes in the alpha layer- which is mentioned as taking each bodypart out of the baking process. And even with those enabled you still throw shadow, and also the textured 1000 remains, which shows the geometry is still there, and just rendered as transparent.

Im curious if that will change when they have the sever side baking included.

[Danny Nolan]

The fact that this isn't live, is kind of silly.

It's not going to 100% morph with the avatar mesh. Alpha layers will still be required. This isn't much of a problem in my eyes. The whole point of the deformer project as to allow you to put on one attachment and have it stick to your av regardless of shape.

Mesh deformer does this.

The delay upon calculating the weights isn't really an issue in my mind. It's a 2-3, 5 second delay. That's not really a reason to just .. not release it. People can make less poly garments, or the customers can wait a few ... seconds for it to deform lol.

I think the reality here is just the tension between Qarl and LL. He made a pretty valiant effort to help them out, and Linden Labs, for some reason refuses to be helped. He even offered to give them the align tool, for free as I recall, but they still refused lol.

[Laura Demme]

I know I'm late to this thread, but considering the topic's importance to so many, I don't think this qualifies as "necroposting."

I've been following this project from the very beginning (/me grins mischievously but good-naturedly at Nalates), and, while I'm not a programmer, I would like to throw my two cents into the discussion.

It was mentioned earlier in this thread that there are many who don't even like mesh, which is true. But I'd be willing to bet that the majority of those who don't like it feel that way because of the currently implemented version of mesh. I feel the deformer would win most of those people over.

As has also been stated, the deformer is as ready as it's going to be. It should be released. If the Lab has reservations about releasing it, they should make those concerns known. Their continued silence isn't helping anything. If they open up and people complain about their reasons...well some people are going to complain no matter what they do. If anyone expects the community to ever be 100% satisfied, that's just not realistic. If the project is simply waiting for people to become available to do the implementation work, then please just TELL US!

As a clothing designer, I've held back from mesh solely because of the current implementation, and I know I'm not the only one. As a clothing buyer, the standard sizing system is ridiculous, mostly because it's not "standard" at all. There is considerable variation in sizing from designer to designer. My avatar is "realistically" scaled and I have to wear sizes ranging from XS to XL, depending on the garment/designer. I've even run into XL items that are still too small.

I know the fashion community is not the entirety of SL, but it IS an awfully large part of the economy. I find it really hard to believe that releasing the deformer would cause more trouble than it's worth. If there's a downside to releasing it now, I'm not seeing it, and from what I can tell, the many programmers and developers that have worked on this project (and who are FAR more knowledgeable than I) aren't seeing it either. I don't feel we should have to speculate.

But if it would cause problems, the Lab should tell us what and/or why, especially given how much time, work AND money the community has put into this project. Are they obligated to tell us anything? No, but it's the right thing to do.

[Rahkis Andel]

I agree with everything you said except that I believe that most of the mesh hate is based around misinformation (Or misplaced expectations) rather than any intrinsic failure of the current implementation of mesh. Mesh is already a capable replacement for a great many prim and sculpt based assets.

What failure does mesh really have currently that isn't already status quo in standard prims or sculpts? Can you create a prim or sculpt clothing asset that conforms perfectly to the avatar? No -- why is mesh expected to perform this way without the deformer?

Currently, the only significant down-side to mesh compared to prims (That I know of) are that you can't make flexi mesh. For what it's worth, I don't really find that to be much of a loss.

Many are just plain misled and sadly it's not only the non-technical types that fall for it. Mesh is blindly blamed with conviction for everything from graphical glitches to skyrocketing LI; In reality, mesh excels in both looking consistently better and having lower LI. Prims? Sculpties? Let's face it, these are hacks. Polygon modeling is how it's been done for time immemorial.

People will learn; I doubt it will be the deformer that changes their minds, however. At least not immediately.

[Laura Demme]

Hi Rahkis. Just to clarify, the mesh dislike I was refering to related only to some members of the fashion-oriented community, specifically regarding garment fit. And you're absolutely right about the superiority of mesh re: looks and land impact. I didn't even realize there were people that didn't care for mesh in other uses. I can't imagine why anyone wouldn't see the benefit. As you said, misinformation must be the reason there. I own an absolutely gorgeous living room set that is only 9LI for 4 detailed pieces. Most of the furnishings and decor of my flat are now mesh, and I think it's fantastic!

[Rahkis Andel]

A few loud voices can drown out the majority; maybe I've paid too much attention to the few loud voices. As a 3d artist I'm probably suceptible to subconciously inflating the degree of opposition there is to a subject I feel strongly about. I sure hope so!

Cheers.

[WhiteRabbit0]

The intelligent useage of UV maps and custom textures ( now with the ... dear god here I go believing that we might actually get it.... normal maps and specular maps ) can take something that would have a 250K poly count if done in sculpts and prims and knock it down to 5K.  It's true, expertise is required, but that's the magnitude of gain that there is potentially.

And to reply to the post above stating that "prims and sculpts don't conform", They are built as clothing elements that way, rather than full clothes.  90% of SL Mesh is a full garment, if your arms are too big you can't just scale the sleeves up.

[Rahkis Andel]

Yes, with rigged mesh, we need the deformer. No doubt.

All I was saying was that mesh itself (unrigged) is no different than a prim or sculpt.

[Tenly]

I agree. LLwill releast this when its ready. And I also noted with the video that while them mesh may bounce with the boobs, it tears in the read and abdomen when the avatar moves. I dont want my clothes doing that. And I think if Linden released the deformer with that kind of defect there would be a backlash accusing them of doing things half-<bleep>.

also noted there is a problem with dae files crashing in the beta. so why release the ability to wear deformer mesh if creators can't even upload their files.

In addition Linden will have to support current Mesh. So, no, I don't agree that Linden is sitting on this for the sake of sitting on it.

its also not our place to dictate or prioritize Linden labs projects.

Linden labs is not a fast food restaurant here to deliver to your demands on your terms.

I say, lets all be patient. Though I can barely wait. There are so many outfits I would wear if only the mesh creators avatar was anything like mine.  

I confess, I want to learn to create with mesh. But Im put off by all the sizes one has to provide...which even with "standards" don't fit everyone.

SL shouldnt be about forcing everyone to fit a standard...it should be about being who you want to be..body and voice. So, Im looking forward to deformer mesh. BUT Id rather have it WHEN Linden is ready.

I can wait for a good thing. Just hoping the waiting will be months rather than years.

my only complaint with SL is the cost of owning your own sim. Especially when you can get a entire for so much cheaper elsewhere. BUT that's a whole other enchiladah.

Tenly 

[Tenly]

I'd like to add one more thing...in order to see this project move forward it takes a lot of people to test and report any issues. Linden has made this easy not only by allowing the download of a beta to test but also with clothing that you can test. see: http://wiki.secondlife.com/wiki/Downloading_test_builds

[Medhue Simoni]

WhiteRabbit, why did you put this in the Mesh forum? It's all developers here, which is why there is more argument than what you were seeking. You should have created this in the General forum.

I agree completely with you. LL clearly doesn't understand the importance of clothing just fitting you, and not making the customer pissed off. The last part is obvious. SL is difficult enough to understand and figure out, the last thing you want is for them to be frustrated all along the way. Buying a dress and putting it on should be a 2 step process, not a 50 step process trying to find which fits, and then remembering which that is, and it is still not perfect. Guess what, she just threw in the towel. I'm a guy, and I won't deal with mesh clothing anymore until this is implemented.

Almost all the arguments against implementing the deformer are from people that aren't that experienced with anything like this. I have every bit of confidense that all the clothing creators will love the deformer and so will their customers. Will they need to rethink some items, or make some items not used by the deformer, well of course, those decisions are part of the creative process. Any mesh clothing creator that doesn't use the deformer, I guarantee will lose all their customers. That's how confident I am about the importance of it.

What I don't quite understand is why it isn't in Firestorm right now. Yeah, ok, LL implements some ridiculous shared experience rule. So what! It's obviously a dumb move, and the deformer project proves that. If not for that rule, it would already be in Firestorm, and every1 would be using it. LL would get a good look at how it performs, without doing any work at all, and not affecting their viewer at all. Personally, I don't even understand how LL even has the power to enforce the rule. Firestorm has more users. Battle over! Firestorm can do whatever it wants! Time and time again, a 3rd party viewer shows LL the way. They obviously shouldn't be the 1 dictating things. It should be the customers, which Firestorm has it's eye on, which is why they have more users.

I'm not a LL basher or a Firestorm basher, or an any1 else basher. I just try to see things without rose colored glasses. Many of the changes LL had to make to the viewer after V2 was not Firestorm's or Pheonix's ideas, they came from LL in the original viewer. LL just tends to lose their focus. Heck, LL still hasn't put the pie menu back into their viewer. This should show how clueless they are sometimes. It only makes sense too, as Lindens come and go, but we, the creators and users in SL stick around.

[Cathy Foil]

I hear you Tenly but the deformer doesn't need anymore testing.  Its limitations are well known and can't really be improved upon unless LL gets a new default avatar mesh which is not going to happen.  If LL is saying it needs more testing that is just a smoke screen to keep us off their backs.

I don't know why LL is sitting on their hands with this project.  Perhaps they are short handed and all resources are being put into implementing the up coming changes to the viewer and servers.  Perhaps they are spiteful and don't want any huge improvements to the SL experience to be attributed to Karl Stiefvater a former  LL employee who's name was Qarl Linden and who when they stupidly laid off caused a wave of protests by many residents.

Just incase you didn't know Karl Stiefvater created the deformer who was paid by our fellow SL residents to create it not LL.

Karl Stiefvater was also responsible for creating sculpties which in my opinion saved SL long enough to get us here before mesh was available.

The deformer is ready and should have been implemented months and months ago.  All that is needed is a few wiki pages on how best to create clothing to work with the deformer and perhaps an instructional video or two.

[Medhue Simoni]

Hey Cathy!

I don't think LL is dragging their feet because Qarl made the deformer. I think it is a number of concerns, all of which I don't think are all that important, at least compared to clothing just fitting you. Some of the concerns may have to do with the creators of clothing, and implementing this and it possibly changing down the road. Again, I'd blow off this concern because clothing is a unique market. 1 whole section of the market is all about what is new. Yes, there are older products that aren't updated, but I don't see where these are damaged by a deformer. It's up to the creator to decide to use the deformer or not. Plus, compared to all the work it take to create the garmet initially, modifications to make it work with the deformer are simple. If LL is concerned about the cost to clothing creators reuploading their garmet again, they should have thought of that before they released mesh. Those creators are currently uploading their garmets dozens of unneccessary times already. If I was a clothing creator, the deformer is a dream improvement as it makes their job dozens of times faster. If I was an active clothing designer, I wouldn't mind reuploading because of a major improvement, as that major improvement will ignite a boom in sales all over again. It's worth every bit of work to reupload the garmet and give your customers a more pleasant clothing experience.

All that said, constant having tiny improvements implemented on a regular basis in other markets, could really annoy creators in other market, but I really do think the clothing market is unique. Plus, I think you have to ask yourself, what are SL's big advantages over other 3D environments or games? I'd say it is the customization of the AVATAR. Now, people can complain all they want about the base avatar, but there is no denying that the customization of it goes far far beyond any character in any game, ever made. So, this is SL's strength. No deformer hurts this strength. Neglecting the animation system hurts this also. Now, IMHO, LL's main core focus should always be to ensure that their avatar is maintained and improved, when possible.

I am not 1 of the people that are calling for a new Avatar 2.0. That, to me, right now, is complete craziness. Part of the big success of SL, was that it gave creators a standard avatar to work with. Each creator was not putting out a totally unique avatar that was not compatible with anything else. It was the compatibility that made creating for that avatar profitable enough to continue to do it. I'm not saying that totally unique mesh avatars are the wrong way to go for a creator. I'm just saying that SL would not have grown the way it has without that standard avatar that every1 could create for. SL grew quickly because every1 was focusing on 1 avatar, and making, literally, millions of products for it. To put out a brand new Avatar 2.0, would be suicide to all the original avatar products, and force creators to do or die. There would be a long transition of awkwardness. The work to implement it would be quite the headache, and a massive cost. In the end, we already have mesh, and that basically takes over, and is better than Avatar 2.0. It would be much more efficient for LL to continue improvements on mesh avatars, than trying to create an Avatar 2.0. If LL just improves mesh avatar features, a few beautiful mesh avatars will become popular, dictated by the consumers, and other creators will start making items specifically for those avatars.

The last point I want to make about Avatar 2.0, is that we already know that the avatar is a major source of lag. I don't see how putting out another avatar with more polygons is going to help with lag, especially considering creators already go overboard with polys for clothing.

I'm not really talking to you Cathy, mostly to all the Avatar 2.0 people.

[Coby Foden]

---

Medhue Simoni wrote:

 

... we already have mesh, and that basically takes over, and is better than Avatar 2.0. It would be much more efficient for LL to continue improvements on mesh avatars, than trying to create an Avatar 2.0. If LL just improves mesh avatar features, a few beautiful mesh avatars will become popular, dictated by the consumers, and other creators will start making items specifically for those avatars.

The last point I want to make about Avatar 2.0, is that we already know that the avatar is a major source of lag. I don't see how putting out another avatar with more polygons is going to help with lag, especially considering creators already go overboard with polys for clothing.

---

I'm wondering:

• You say that mesh [avatars] would be better than avatar 2.0 would be.

• To be better mesh avatars would then need more polygons than default avatar 2.0. (How else they could be better?)

• Then, concerning lag, how could the "better" mesh avatars be less laggy that default avatar 2.0 would be?

 

Then there is the problem with mesh avatars, it would most likely turn out to be so that each mesh avatar would need their own specific skins and clothings, not being compatible with each other - more headaches for clothing designers and consumers alike.  You pointed out that the standard avatar is big part of the success of SL - indeed it is so.  With mesh avatars there is no standard.  With default avatar 2.0 there would be the standard again.

I would like see that Linden Lab would finally take the step and implement better default avatar.  It could even be possible to make it so that present default avatar and the new default avatar would co-exist.  Surely it would add complexity for the viewer and server codes, but I think it would not be impossible to make it so.

Over longish period of time the present default avatar would die naturally out of old age as less and less people would be using it, and less and less designers would create accessories for the old avatar.

Even though the present avatar is highly customizable, everybody knows how very bad its mesh is.  As an animator you surely know how easily the mesh breaks even in some slight animations in some places.  It surely would be the time to upgrade the avatar mesh.  Or should we perhaps be satisfied with it forever - as long as SL exist?  My thinking is that mesh avatars are not the solution for the underlying problem.